RBT Data Collection and Graphing Study Guide — Unit A (2026 Task List)

Content reviewed by James Fuller, BCBA · Last updated July 19, 2026 · Aligned to BACB RBT Test Content Outline (3rd Edition)

Quick Answer

Data Collection and Graphing (Unit A) accounts for 13 out of 75 scored questions (17%) on the RBT exam — the third-largest domain. It covers 8 tasks: continuous measurement (frequency, duration, latency, interresponse time), discontinuous measurement (partial interval, whole interval, momentary time sampling), permanent product recording, entering data and updating graphs, describing behavior in observable and measurable terms, calculating and summarizing data (rate, mean duration, percentage), identifying trends in graphed data, and describing the risks of unreliable data collection and poor procedural fidelity.

Unit A is the foundation the rest of the RBT practice exam sits on. Every behavior plan you’ll ever implement depends on someone measuring behavior accurately — and in almost every case, that someone is you. Get measurement wrong and every clinical decision downstream is built on bad information.

This is also the only unit with actual arithmetic. Rate, percentage, mean duration — you’ll need to calculate all three, and the exam expects you to do it without a calculator. The math itself is simple. The trap is knowing which calculation a question is asking for. To see where you stand, take a full-length 85-question practice exam after working through this guide, or browse every study resource at our free RBT practice test hub.

This guide covers all 8 tasks (A.1 through A.8) in the exact order the BACB lists them, with worked calculations, measurement comparison tables, and the specific errors that cost candidates points. Everything is reviewed by James Fuller, BCBA.

One idea worth holding onto before you start: measurement methods aren’t interchangeable. Each one answers a different question about behavior. Frequency tells you how often. Duration tells you how long. Latency tells you how quickly it started. Choosing the wrong method doesn’t give you slightly worse data — it gives you data about the wrong thing entirely.

Quick Facts About Unit A — Data Collection and Graphing

• Number of scored questions: 13 out of 75 (17% of the exam)
• Total tasks in this domain: 8 (numbered A.1 through A.8)
• Effective Test Content Outline: 3rd Edition (in effect for 2026 exams)
• Recommended study time: 8-12 hours
• Difficulty level: Moderate to high — the only unit requiring calculations
• Core skills tested: frequency, duration, latency, IRT, interval recording, momentary time sampling, permanent product, rate and percentage calculations, graph reading, procedural fidelity
• Most common exam trap: Confusing latency with interresponse time, and partial with whole interval recording

What Is Data Collection in ABA?

Data collection in Applied Behavior Analysis is the systematic recording of behavior using objective, quantifiable methods. It’s what separates ABA from opinion-based intervention — instead of asking whether a client “seems better,” you look at whether the numbers moved.

Every measurement method falls into one of three broad categories:

• Continuous measurement — every instance of the behavior is recorded (A.1)
• Discontinuous measurement — behavior is sampled during intervals rather than recorded continuously (A.2)
• Permanent product — the lasting outcome of the behavior is measured after the fact (A.3)

The Central Trade-Off

Continuous measurement is more accurate but requires your full attention. Discontinuous measurement is less accurate but lets you teach while collecting data. The BCBA chooses based on that trade-off — accuracy versus practicality — and the exam frequently asks you to recognize which method fits a given situation.

Your role as an RBT is to implement the measurement method specified in the plan exactly as written, record accurately in real time whenever possible, and report any problems with the data to your supervising BCBA. You don’t choose the measurement method and you don’t switch methods because another one would be easier.

The 8 Tasks in Unit A (Per BACB 3rd Edition)

The BACB Test Content Outline (3rd Edition) organizes Data Collection and Graphing into 8 tasks.

1. A.1 — Implement continuous measurement procedures (e.g., frequency, duration, latency, interresponse time)
2. A.2 — Implement discontinuous measurement procedures (e.g., partial & whole interval, momentary time sampling)
3. A.3 — Implement permanent product recording procedures
4. A.4 — Enter data and update graphs
5. A.5 — Describe behavior and environment in observable and measurable terms
6. A.6 — Calculate and summarize data in different ways (e.g., rate, mean duration, percentage)
7. A.7 — Identify trends in graphed data
8. A.8 — Describe the risks associated with unreliable data collection and poor procedural fidelity

Continuous Measurement Procedures (A.1)

Task A.1 covers “continuous measurement procedures (e.g., frequency, duration, latency, interresponse time).” Continuous measurement captures every instance of the behavior — nothing is sampled or estimated.

Frequency (count)

A simple tally of how many times the behavior occurred.

Use when: The behavior has a clear beginning and end, and each instance is roughly the same length.

Example: The client raised their hand 7 times during the session.

Don’t use when: Instances vary widely in length. Three tantrums lasting 30 seconds and three lasting 20 minutes both count as “3” — which tells you almost nothing useful.

Rate

Frequency divided by time. Rate lets you compare across sessions of different lengths, which raw frequency can’t do.

Example: 12 occurrences in a 30-minute session equals a rate of 0.4 per minute.

Duration

How long the behavior lasts, from onset to offset.

Use when: The length of the behavior is what matters clinically.

Example: The tantrum lasted 8 minutes and 20 seconds. Or across a session: total duration of on-task behavior was 22 minutes.

Latency

The time between the stimulus or instruction and the start of the behavior.

Use when: You care about responsiveness — how quickly the learner initiates after being asked.

Example: The RBT said “start your worksheet” at 10:02:00. The client picked up the pencil at 10:02:45. Latency = 45 seconds.

Interresponse Time (IRT)

The time between the end of one response and the start of the next response. Behavior-to-behavior, with no instruction involved.

Use when: You want to know how spaced out or clustered responses are.

Example: The client hit at 10:05:00 and again at 10:07:30. IRT = 2 minutes 30 seconds.

The #1 Unit A Exam Trap

Latency and IRT are constantly confused. The distinction is what starts the clock. Latency is measured from an instruction or stimulus to the behavior. IRT is measured from one behavior to the next behavior. If the scenario mentions a demand, instruction, or SD, it’s latency. If it only mentions two occurrences of the same behavior, it’s IRT.

Continuous measurement comparison

Discontinuous Measurement Procedures (A.2)

Task A.2 covers “discontinuous measurement procedures (e.g., partial & whole interval, momentary time sampling).” These methods divide the observation period into intervals and sample behavior rather than recording every instance.

Discontinuous methods exist for one practical reason: you often need to teach and collect data at the same time. Continuous measurement demands your full attention. Interval methods let you do both, at the cost of some accuracy.

Partial interval recording

Score the interval if the behavior occurred at any point during it — even for one second.

Example: A 10-minute observation divided into 20 intervals of 30 seconds. If screaming happened at any moment within an interval, that interval is scored.

Bias: overestimates. A behavior that occurred for 2 seconds in a 30-second interval scores the whole interval, making the behavior look more prevalent than it was.

Best used for: Behaviors you want to decrease. The overestimation bias is conservative in that direction — it won’t let you conclude a problem behavior improved when it hasn’t.

Whole interval recording

Score the interval only if the behavior occurred for the entire interval, start to finish.

Example: If on-task behavior must be continuous through all 30 seconds to score the interval, a single 2-second glance away means the interval isn’t scored.

Bias: underestimates. Any brief interruption disqualifies the interval, making the behavior look less prevalent than it was.

Best used for: Behaviors you want to increase. The underestimation bias won’t let you conclude a skill improved when it hasn’t.

Momentary time sampling (MTS)

Score the interval based on whether the behavior is occurring at the exact moment the interval ends. What happens during the rest of the interval is irrelevant.

Example: A timer sounds every 5 minutes. At each signal you look up and record whether the client is on-task at that instant.

Bias: least biased of the three. MTS doesn’t systematically over- or underestimate, which makes it the most accurate discontinuous method for estimating duration.

Best used for: Continuous, longer-duration behaviors like on-task behavior or stereotypy, especially when you need to teach simultaneously or observe a group.

Discontinuous measurement comparison

Memory Hook

Partial = Part of the interval. Whole = Whole interval. MTS = the Moment the timer sounds. Then pair each with its bias: partial over, whole under, MTS neither.

Permanent Product Recording (A.3)

Task A.3 covers permanent product recording — measuring the tangible outcome the behavior leaves behind, rather than observing the behavior itself.

How it works

You don’t need to watch the client work. You count the result afterward.

• Number of math problems completed on a worksheet
• Number of toys returned to the bin after cleanup
• Number of words written in a journal entry
• Number of dishes washed and placed in the rack

Why it’s useful

• No real-time observation required
• The record is permanent and can be re-verified later
• Highly accurate for behaviors that produce countable outcomes
• Frees the RBT to teach during the activity

The main limitation

Permanent product tells you the outcome, not the process. If a worksheet has 15 completed problems, you know 15 problems were completed — but not by whom, in what order, or with how much prompting. If a sibling helped, the data is wrong and you’d never know from the product alone.

Permanent product also can’t capture behaviors that leave no trace — vocal stereotypy, hand flapping, or eye contact produce nothing to count later.

Entering Data and Updating Graphs (A.4)

Task A.4 covers entering data and updating graphs. Line graphs are the standard display in ABA because they show change over time clearly.

The parts of a line graph

• X-axis (horizontal): Time — sessions, days, or dates
• Y-axis (vertical): The behavior measure — frequency, rate, percentage, duration
• Data points: One point per session
• Data path: Lines connecting consecutive data points
• Phase change line: A vertical line marking a condition change, such as baseline to intervention
• Phase labels: Text identifying each condition
• Axis labels: What each axis represents, including units

Graphing conventions that get tested

1. Never connect data points across a phase change line. The data path breaks at each phase change. This is the most frequently tested graphing rule.
2. Don’t connect across missing sessions. If a session was missed, leave a gap rather than drawing a line across it.
3. Keep the Y-axis scale consistent within a graph. Changing the scale mid-graph distorts the visual.
4. Enter data promptly — same day where possible.
5. Label everything. A graph should be interpretable without additional explanation.

Exam Trap

Questions often describe a graph where the data path continues straight through the phase change line and ask what’s wrong. The answer is that data paths must break at phase changes — connecting across them implies continuity between conditions that doesn’t exist.

Describing Behavior in Observable and Measurable Terms (A.5)

Task A.5 requires describing behavior and environment in observable and measurable terms. This is the operational definition — and it’s the prerequisite for every other task in this unit. You cannot measure what you haven’t defined precisely.

The three criteria for a good operational definition

1. Observable — you can see or hear it. Internal states don’t qualify.
2. Measurable — you can count it, time it, or otherwise quantify it.
3. Clear enough for agreement — two independent observers watching the same session would record the same thing.

Poor versus operational definitions

Notice what the operational versions share: specific actions, specific thresholds, and specific time windows. Nothing is left to the observer’s judgment.

The dead man’s test

A useful check: if a dead man could do it, it isn’t behavior. “Sitting quietly,” “not hitting,” and “remaining still” all fail the test — they describe the absence of behavior rather than behavior itself. Define what the learner does, not what they don’t do.

Calculating and Summarizing Data (A.6)

Task A.6 covers calculating and summarizing data “in different ways (e.g., rate, mean duration, percentage).” These are the three calculations the exam expects you to perform.

Rate

Formula: Rate = Total count ÷ Total time

Worked example: A client engaged in 24 instances of hand-raising during a 60-minute session.

24 ÷ 60 = 0.4 responses per minute

Second example: 15 occurrences in a 45-minute session.

15 ÷ 45 = 0.33 per minute (or 20 per hour)

Why rate matters: A count of 15 means something completely different in a 10-minute session than in a 3-hour session. Rate makes sessions comparable.

Percentage

Formula: Percentage = (Number of occurrences ÷ Total opportunities) × 100

Worked example: The client responded correctly on 17 of 20 trials.

(17 ÷ 20) × 100 = 85%

Second example: The client complied with 9 of 12 instructions.

(9 ÷ 12) × 100 = 75%

For interval data: Percentage of intervals = (Intervals scored ÷ Total intervals) × 100. If 14 of 40 intervals were scored: (14 ÷ 40) × 100 = 35% of intervals.

Mean duration

Formula: Mean duration = Total duration ÷ Number of occurrences

Worked example: Four tantrums lasting 3, 7, 5, and 9 minutes.

Total = 3 + 7 + 5 + 9 = 24 minutes
24 ÷ 4 = 6 minutes mean duration

Mean latency

Same structure as mean duration, applied to latency values.

Worked example: Latencies of 12, 8, 20, and 16 seconds across four instructions.

Total = 56 seconds
56 ÷ 4 = 14 seconds mean latency

Calculation summary

Read the Question Carefully

Most calculation errors on the exam aren’t arithmetic errors — they’re using the wrong denominator. Percentage of trials uses total trials. Percentage of intervals uses total intervals. Mean duration divides by number of occurrences, not by session length. Identify what’s being asked before you calculate.

Identifying Trends in Graphed Data (A.7)

Task A.7 covers identifying trends in graphed data. You won’t be asked to make clinical decisions from graphs — that’s the BCBA’s job — but you must be able to describe what a graph shows.

Trend

The overall direction of the data path across time.

• Increasing (ascending) trend — data points move upward over sessions
• Decreasing (descending) trend — data points move downward over sessions
• Zero trend (flat) — no consistent directional change

Direction alone doesn’t tell you whether things are going well. An increasing trend is good for a skill acquisition target and bad for a problem behavior. Always check what’s on the Y-axis.

Level

The average value of the data within a phase — roughly, how high or low the data sits on the Y-axis. Comparing level across phases shows whether the intervention shifted the behavior’s magnitude.

Variability

How much the data points scatter around the trend.

• Low variability (stable): Points cluster tightly. Patterns are easy to interpret.
• High variability: Points scatter widely. Harder to interpret, and often a signal that something inconsistent is happening — inconsistent implementation, changing setting events, or an unstable environment.

High variability during baseline is a common reason a BCBA extends baseline before starting intervention.

Reading a graph in four steps

1. Check the axes. What’s being measured, and over what time period?
2. Find the phase change lines. How many conditions are shown?
3. Describe the trend in each phase — increasing, decreasing, or flat.
4. Compare level and variability across phases. Did the behavior shift, and is the data stable enough to trust?

What RBTs do and don’t do with graphs

You update graphs, describe what they show, and flag anything unusual to your BCBA. You do not interpret clinical significance, decide whether the intervention is working, or recommend changes based on the graph.

Risks of Unreliable Data Collection and Poor Procedural Fidelity (A.8)

Task A.8 asks you to describe the risks associated with unreliable data collection and poor procedural fidelity. This task is unusual — it tests understanding of consequences rather than execution of a procedure.

What procedural fidelity means

Procedural fidelity, also called treatment integrity, is the degree to which a procedure is implemented exactly as written. High fidelity means you did what the plan said. Low fidelity means you deviated — whether you meant to or not.

Why unreliable data is dangerous

The BCBA can only make decisions from what’s recorded. When the data is wrong, every decision built on it is wrong too:

• An effective intervention gets discontinued because inaccurate data made it look ineffective
• An ineffective intervention continues because inflated data made it look successful
• Treatment time is wasted that the client can’t get back
• Problem behavior escalates while an unhelpful plan stays in place
• Documentation and billing problems arise when records don’t reflect what happened

Why poor procedural fidelity is dangerous

When a procedure isn’t implemented as written, you can’t tell whether the plan failed or the implementation did.

• The BCBA may revise a plan that would have worked if run correctly
• Inconsistency across staff confuses the learner and slows progress
• Results can’t be attributed to the intervention with any confidence
• In behavior reduction, inconsistent implementation can make behavior worse — intermittently reinforced behavior is more resistant to change than continuously reinforced behavior

Common sources of measurement error

• Recording from memory at the end of a session instead of in real time
• Unclear operational definitions leaving room for observer judgment
• Observer drift — your definition gradually loosening over weeks without you noticing
• Missing occurrences while attending to other tasks
• Recording what you expect to see rather than what occurred
• Using a measurement method that doesn’t fit the behavior

Interobserver agreement (IOA)

IOA is the check on data reliability — two observers independently record the same behavior and their records are compared. High agreement suggests the definition is clear and observers are applying it consistently. Low agreement signals that the definition needs sharpening or observers need retraining.

How to protect data quality

1. Record in real time, not from memory
2. Know the operational definition precisely before the session starts
3. Ask for clarification when a definition is ambiguous
4. Use the measurement method specified in the plan, without substituting
5. Report honestly, including sessions that went badly
6. Report data irregularities to your BCBA rather than smoothing them over

Exam Trap

Scenarios sometimes describe an RBT who missed recording several occurrences and ask what to do. The correct answer is to record what was actually observed and report the gap to the BCBA — never to estimate, fill in from memory, or approximate the missing data.

What RBTs Can and Cannot Do in Data Collection

What RBTs do

• Implement the measurement method specified in the plan
• Record data accurately and in real time
• Enter data and update graphs as directed
• Describe what graphed data shows
• Report data irregularities and measurement problems to the BCBA
• Request clarification on unclear operational definitions
• Participate in IOA sessions when scheduled

What RBTs do not do

• Select or change the measurement method
• Write or revise operational definitions
• Interpret graphs to determine whether an intervention is working
• Make clinical decisions based on data trends
• Estimate or reconstruct missing data
• Modify recorded data after the fact

How to Study Unit A Efficiently

Unit A is 17% of the exam and rewards drilling more than reading. The concepts are straightforward; the errors come from confusion between similar methods and from misreading calculation questions.

Step 1: Lock down latency versus IRT first

This single distinction generates a disproportionate number of missed questions. Latency starts at an instruction. IRT starts at the end of a behavior. Write five scenarios for each and label them until the answer is instant.

Step 2: Memorize the interval biases

Partial overestimates, whole underestimates, MTS is least biased. Then connect each to its use case — partial for behaviors you’re decreasing, whole for behaviors you’re increasing. Exam questions test the pairing, not just the definition.

Step 3: Practice all three calculations by hand

Work rate, percentage, and mean duration problems without a calculator until the formulas are automatic. Then practice identifying which formula a word problem is asking for — that’s where most errors actually happen.

Step 4: Take the topic quiz until you score 90%+

The Data Collection and Graphing quiz covers only Unit A content. Take it repeatedly and read every explanation, including for questions you answered correctly.

Step 5: Apply Unit A on a full-length exam

Calculation questions are the easiest to rush when time pressure builds. Practice them under realistic conditions with a complete practice exam with two versions — 170 total questions across all six domains — and track your accuracy on Unit A items specifically.

Seven Common Exam Traps in Unit A

Trap 1: Confusing latency with interresponse time

Latency is measured from an instruction or stimulus to the behavior. IRT is measured from one behavior to the next. Look for whether a demand or SD is mentioned in the scenario.

Trap 2: Mixing up partial and whole interval

Partial scores the interval if the behavior occurred at any point. Whole requires the behavior throughout the entire interval. Partial overestimates; whole underestimates.

Trap 3: Misidentifying momentary time sampling

MTS records only what’s happening at the exact moment the interval ends. Behavior during the rest of the interval is irrelevant.

Trap 4: Using the wrong denominator in a calculation

Percentage of trials uses total trials. Percentage of intervals uses total intervals. Mean duration divides by number of occurrences, not session length.

Trap 5: Connecting data across phase change lines

Data paths must break at every phase change. Connecting across implies continuity between conditions that doesn’t exist.

Trap 6: Estimating missing data

If occurrences were missed, record what was observed and report the gap. Never estimate, reconstruct from memory, or approximate.

Trap 7: Assuming an increasing trend is good

Direction alone means nothing without knowing what’s being measured. An increasing trend in aggression is bad; an increasing trend in independent responding is good. Check the Y-axis first.

Frequently Asked Questions About RBT Data Collection and Graphing

How many Data Collection and Graphing questions are on the RBT exam?

The RBT exam includes 13 Data Collection and Graphing questions out of 75 scored items — 17% of the exam. Unit A is the third-largest domain.

How many tasks are in Unit A?

The BACB RBT Test Content Outline (3rd Edition) organizes Unit A into 8 tasks numbered A.1 through A.8, covering continuous measurement, discontinuous measurement, permanent product recording, entering data and updating graphs, operational definitions, data calculations, identifying trends, and the risks of unreliable data.

What is the difference between continuous and discontinuous measurement?

Continuous measurement records every instance of the behavior, producing more accurate data but requiring full attention. Discontinuous measurement samples behavior during intervals, which is less accurate but allows the RBT to teach while collecting data.

What is the difference between latency and interresponse time?

Latency is the time from an instruction or stimulus to the start of the behavior. Interresponse time (IRT) is the time between the end of one response and the start of the next response. If a demand or SD is involved, it is latency. If only two occurrences of the behavior are involved, it is IRT.

What is partial interval recording?

Partial interval recording scores an interval if the behavior occurred at any point during it, even briefly. It overestimates the true occurrence of behavior, which makes it appropriate for behaviors targeted for decrease.

What is whole interval recording?

Whole interval recording scores an interval only if the behavior occurred throughout the entire interval. It underestimates the true occurrence of behavior, which makes it appropriate for behaviors targeted for increase.

What is momentary time sampling?

Momentary time sampling scores an interval based only on whether the behavior is occurring at the exact moment the interval ends. It is the least biased discontinuous method and works well for continuous behaviors or group observation.

Which interval method overestimates behavior?

Partial interval recording overestimates behavior because a single brief occurrence scores the entire interval. Whole interval recording underestimates because any interruption disqualifies the interval.

How do you calculate rate in ABA?

Rate equals total count divided by total time. For example, 24 occurrences in a 60-minute session equals a rate of 0.4 responses per minute.

How do you calculate percentage of correct responses?

Divide the number of correct responses by the total number of opportunities and multiply by 100. For example, 17 correct out of 20 trials equals 85%.

How do you calculate mean duration?

Divide total duration by the number of occurrences. For example, four tantrums lasting 3, 7, 5, and 9 minutes total 24 minutes; 24 divided by 4 equals a mean duration of 6 minutes.

What is permanent product recording?

Permanent product recording measures the tangible outcome a behavior produces rather than observing the behavior itself — such as counting completed math problems or toys returned to a bin. It does not require real-time observation but cannot verify who produced the outcome or how.

What is an operational definition?

An operational definition describes behavior in observable and measurable terms, specific enough that two independent observers watching the same session would record the same thing. It states what the learner does, not what they feel or intend.

Why should data paths break at phase change lines?

Connecting data across a phase change line implies continuity between two different conditions. Because the conditions differ, the data path must break at each phase change to accurately represent the separate phases.

What is procedural fidelity?

Procedural fidelity, also called treatment integrity, is the degree to which a procedure is implemented exactly as written. Poor fidelity makes it impossible to determine whether an intervention failed or was simply implemented incorrectly.

What are the risks of unreliable data collection?

Unreliable data can lead the BCBA to discontinue an effective intervention, continue an ineffective one, waste treatment time, allow problem behavior to escalate, and create documentation and billing problems.

What is interobserver agreement (IOA)?

IOA is a reliability check in which two observers independently record the same behavior and their data is compared. High agreement suggests the operational definition is clear and applied consistently; low agreement signals the definition needs revision or observers need retraining.

What should an RBT do if they missed recording some occurrences?

Record only what was actually observed and report the gap to the supervising BCBA. Estimating, reconstructing from memory, or approximating missing data compromises the integrity of the record.

How many hours should I study Unit A for the RBT exam?

Most candidates spend 8 to 12 hours on Unit A. The highest-value activities are drilling the latency versus IRT distinction, memorizing the interval recording biases, and practicing rate, percentage, and mean duration calculations by hand.

Practice Data Collection and Graphing Content

The fastest way to lock in Unit A content is active practice with scenario-based questions and calculation problems.

TOPIC QUIZ Data Collection & Graphing Quiz Focused questions covering only Unit A — with detailed explanations for every answer. Take the quiz → FULL-LENGTH EXAM 85-Question Mock Exam Full-length practice exam with Version 1 and Version 2 — 170 total unique questions. Take the exam →

Other RBT Study Guide Units

• Unit B: Behavior Assessment (8 questions, 11%)
• Unit C: Behavior Acquisition (19 questions, 25%)
• Unit D: Behavior Reduction (14 questions, 19%)
• Unit E: Documentation and Reporting (10 questions, 13%)
• Unit F: Ethics (11 questions, 15%)
• Return to the full RBT Study Guide

About the BCBA Reviewer

All content in this Data Collection and Graphing study guide is reviewed by James Fuller, BCBA, a Board Certified Behavior Analyst with clinical experience supervising RBTs across Applied Behavior Analysis service settings. James reviews study guide content for clinical accuracy, alignment with the current BACB RBT Test Content Outline (3rd Edition), and consistency with current examination policies.

James Fuller’s certification can be verified through the official BACB Certificant Registry. Content on this page was most recently reviewed on July 19, 2026.

This site is independently created and operates without affiliation to the Behavior Analyst Certification Board (BACB) or Pearson VUE. For official RBT examination policies, refer to the BACB website and the official RBT Test Content Outline (3rd Edition).

Practice More

For full preparation, explore these: