Control, ethics, reliability and validity

Introduction to Quality Control in Psychology

Welcome! In this chapter, we are looking at the "Quality Control" department of Psychology. When researchers conduct a study, they don't just want to find an answer; they want to find an answer that is accurate, consistent, and fair. This chapter covers the tools psychologists use to make sure their results are trustworthy and that their participants are treated with respect. Don't worry if these terms seem a bit dry at first—once you see them as the "rules of the game," they become much easier to master!

1. Control: Dealing with Variables

In an experiment, we change the Independent Variable (IV) to see the effect on the Dependent Variable (DV). However, other "nuisance" variables can get in the way. We need to control these to make sure our results are "clean."

Extraneous and Confounding Variables

Extraneous Variables: These are any variables other than the IV that might affect the DV. They are like "background noise." If they aren't controlled, they make it harder to see the real effect of the IV. Examples include the temperature of the room or the time of day.

Confounding Variables: These are the "troublemakers." A confounding variable is an extraneous variable that did vary systematically with the IV. This means we can't tell if the change in the DV was caused by the IV or the confounding variable. It ruins the experiment's internal validity.

Demand Characteristics and Investigator Effects

Sometimes, the "human" element of research causes problems:

Demand Characteristics: This happens when participants pick up on "cues" about the aim of the study. They might change their behavior to help the researcher (the "Please-U effect") or purposefully spoil the results (the "Screw-U effect").

Investigator Effects: This is when the researcher’s own behavior or characteristics (like their tone of voice, body language, or even their gender) unconsciously influence the participants' performance or the way data is recorded.

How do we control these?

Psychologists have several "procedural tricks" to keep things fair:

Standardisation: Ensuring that every participant has the exact same experience. This includes using standardised instructions (reading the same script to everyone).

Randomisation: Using chance (like flipping a coin or using a random number generator) to decide the order of tasks or the presentation of materials. This reduces the researcher's unconscious bias.

Random Allocation: Randomly assigning participants to different groups (e.g., the experimental group vs. the control group) to ensure that participant variables (like intelligence or personality) are spread evenly across the conditions.

Counterbalancing: Used in repeated measures designs to stop order effects. Half the participants do Condition A then B, and the other half do B then A (the \(ABBA\) technique). This ensures that boredom or practice effects don't just affect one condition.

Key Takeaway: Control is all about making sure the IV—and only the IV—is affecting the DV.

2. Ethics: The Rules of Conduct

Psychological research must be ethical. In the UK, we follow the British Psychological Society (BPS) code of ethics. There are four major principles you need to know:

1. Informed Consent: Participants should be told the aims of the research so they can make a settled decision about whether to take part. Tip: If participants are under 16, a parent must provide consent.

2. Deception: This involves deliberately misleading participants or withholding information. While sometimes necessary to avoid demand characteristics, it should be avoided whenever possible.

3. Protection from Harm: Participants should not be placed at any more risk than they would experience in their daily lives. This includes both physical harm and psychological stress (like embarrassment or loss of self-esteem).

4. Privacy and Confidentiality: Participants have the right to control information about themselves. Their data must be protected and usually kept anonymous (using numbers or initials instead of names).

How do we deal with ethical issues?

Debriefing: At the end of a study, participants should be told the true aims, any deception should be revealed, and they should be offered the right to withdraw their data if they are unhappy.

Key Takeaway: Ethics is about balancing the need for scientific knowledge with the rights and well-being of the participants.

3. Reliability: The Consistency Factor

Reliability is simply another word for consistency. If a test is reliable, it should produce the same result every time it is used.

Types of Reliability

Test-retest Reliability: Giving the same participants the same test on two different occasions. If the results are highly similar, the test is reliable. We check this using a correlation coefficient; it should be \(+0.80\) or higher to be considered reliable.

Inter-observer Reliability: In observations, two or more researchers watch the same behavior and record it independently using the same behavioural categories. Their results are then compared. If they agree, the observation is reliable.

How to improve Reliability

Standardise the procedures: Make sure every condition is identical.
Operationalise variables: Define exactly what you are measuring. For example, instead of "aggression," measure "number of times a participant hits the doll."
Train observers: Ensure everyone understands the behavioural categories perfectly.

Key Takeaway: If a measurement is "all over the place," it is unreliable. Reliable tools are consistent tools.

4. Validity: The Truth Factor

Validity is about accuracy. Does the test actually measure what it claims to measure? Does it represent the real world?

Internal vs. External Validity

Internal Validity: Did the IV really cause the change in the DV? Or was it caused by a confounding variable or demand characteristics? If a study is well-controlled, its internal validity is high.

External Validity: Can the findings be generalised beyond the research setting? This includes:

Assessing Validity

Face Validity: A basic check—"on the face of it," does the test look like it measures what it's supposed to?

Concurrent Validity: Comparing the results of a new test with a pre-existing, well-established test of the same thing. If the results are similar (a correlation of \(+0.80\) or above), the new test has concurrent validity.

Improving Validity

Control groups: Use a control group to show that the IV is really what’s causing the effect.
Single-blind/Double-blind procedures: In a single-blind study, the participant doesn't know which condition they are in (reduces demand characteristics). In a double-blind study, neither the participant nor the researcher knows (reduces investigator effects).
Operationalisation: Making sure the way you measure the DV is as precise as possible.

Key Takeaway: Validity is about truth. A test can be reliable (consistent) but not valid (accurate). Imagine a scale that always tells you that you weigh 5kg less than you do—it's reliable because it's consistent, but it's not valid because it's wrong!

Summary Table: Quick Review

Control: Removing "nuisance" variables to ensure a fair test.
Ethics: Following BPS guidelines to protect participants.
Reliability: Making sure the measurement is consistent over time and between researchers.
Validity: Making sure the measurement is accurate and generalisable.

Did you know? Even if a study has very high control (like a lab experiment), it often has lower ecological validity because the environment is so artificial. Research is often a "balancing act" between these two things!