Research methods paper 2 (booklet 2)

Qualitative

word data (anything that is not numerical data

Methods of qualitative data collection

• Interviews 

• Questionnaires (open questions)

• Unstructured observation

Advantages of qualitative data

More detailed information

Disadvantages of qualitative data

• Costs more money- less research funding provided in result 

• Slower and more difficult to analyse

Quantitative data

numerical data

Methods of quantitative data collection

Structured observations (closed questions)

Advantages of quantitative data

• Quicker and easier to analyse 

• Cheaper

Disadvantages of quantitative data

• Less detailed information

Primary data

data you collect yourself from direct observation 

Advantages of primary data

More reliable as you carried out the experiment yourself

Disadvantages of primary data

• Expensive 

• Time consuming

Secondary data

data that has been collected from another research that you are just analysing  

Advantages of secondary data

• Cheaper 

• Faster to obtain

Disadvantages of secondary data

• Potential of investigator bias because it wasn't collected by you

• Less reliable

Nominal

Qualitative values- usually tallied- frequencies, not able to rank 

• E.g. gender, weather, ethnicity, marital statu

Ordinal

scaled or ranked data (ordered). Will be subjective ratings. Often seen as a source. 

• E.g. 1-5 on a likert scale (usually a score)

Interval

ranked data with equal measurement intervals/standardised measurements and units, objective with arbitrary zero. Uses pre-existing measurement scales 

• E.g. time, temperature, bank balance

Ratio

same as interval, but includes an absolute zero 

• E.g. cash, distance, weight

Central tendency

tells us about the most typical values in a data set

The most common measure of central tendency

• Mean

• Mode

• Media

Mean

is the average value, only used in interval and ratio data

Advantages:

• Includes all values in data set so most representative 

Disadvantages:

• Easily affected by outliers

• Only usable on interval level data 

Median

is the middle value, only used in ordinal data 

Advantages:

• Not affected by extreme values 

• Easy to calculate 

Disadvantages:

• Less sensitive than mean so less representative

Mode

is the most frequent value, only used in nominal data

Advantages:

• Easy to calculate 

• Can be sued on all levels of data 

• Only one that can be used on nominal data  

Disadvantages:

• If several modes, then it can’t be used 

• Does Not represent the whole data set

Dispersions

tell us how for scores vary and differ  between one another

The most common measure of dispersion

• Range

• Standard deviation

Range

the spread of data. Difference between highest and lowest value 

Advantages:

• Simple to calculate 

• Can be used on ordinal or interval level data 

Disadvantages:

• Not a clear representative of most of the score 

Standard deviation

 average spread of values around the mean. Interval and ratio

Advantages:

• Clear representation of how far most values are from the mean 

• More precise than range 

Disadvantages:

• Affected by extreme values but this may not be revealed

Data representations

• Graphs and tables

• Bar chart

• histogram

• contingency table

• scatter graph/scattergram

Graphs and tables

• Graphs need to be labelled: title, x-axis, y-axis 

• Tables have subtitles describing table conteny

Bar chart

• Nominal data only (=separated bars)

• Height of bars represent frequencies 

Histogram

• Continuous data- ordinal, interval, ration (no gaps needed between bars)

• Shows frequency of data in successive numerical intervals 

• IV plotted along the x-axis, DV along the bottom 

Contingency table

• Raw scores displayed in columns and rows

• Often asks for you to draw conclusions from the data 

Scatter graph/scattergram

• Gives a good visual picture of the relationship between the two variables 

• Aids the interpretation of the correlation coefficient

What is a distribution?

The overall pattern of data seen in a graph. It is always predicted that data will follow a normal distribution

Negatively skewed distribution

Most of the distribution is concentrated to the right (higher end) and the long tail is to the left (lower end)

Positively skewed distribution

Most of the distribution is concentrated to the left (lower end) and the long tail it to the right (higher end)

Inferential statistics

determining whether a study has significant findings so that you can generalise these findings to the population (outside the confines of the research settings)

Probability

how likely something is to happen 

(p)

• a number between 0 and 1

• 0 meaning impossible and 1 meaning certain

p≤0.05

this means that the likelihood/probability of the behaviour not happening again is equal to or less than 5%- to convey significant findings

• This means that 95 people out of 100 did predict this behaviour 

• This cannot be fluke or coincidence 

• Any higher still counts as significant 

• It is even harder to achieve (e.g. p≤0.01)

Challenging other research

when challenging well known theories, psychologists tend to adopt more stringent significance levels e.g. p≤0.02 or 0.1 (only 1 in 100 fail to do this behaviour)

Proof

• does not exist in psychological research, unless 100% accuracy is found (p=0.00 does not exist). 

• P≤0.05 shows significant support for a theory, but not proof 

Type 1 error

• False positive 

• Belief that a significant difference or correlation is found- but this is an error one does not exist)

• Rejecting the null hypothesis (when we shouldn’t as its actually true)

Type 2 error

• False negative 

• Belief that no significant difference or correlation has been found- but this is an error (there is a significant difference/correlation in the data in reality)

• Accepting a null hypothesis (when we shouldn’t because it's actually false)

Avoiding or reducing the chances of errors

• Making the significant levels stricter, reduces the chance of a type 1 error (e.g. p≤0.01) , but will increase the chance of making a type 2 error 

• To reduce the chance of both error types, simply increase the sample size (=increase in validity also) 

one tailed

directional test

two tailed

non-directional test

Is this test parametric?

1. interval/ratio level data

2. Conforms to a normal distribution (not skewed)

3. Spread of data is not significantly large

When to use the sign test:

• When you have nominal data (non-parametric)

• When you are checking for differences in data (not relationships)

• This test is used when repeated measures experimental design has been used. (i.e. the same participant has 2 results). This means the data is paired or related 

• It can also be used when a  matched pairs design has been used (using the pairs of participants as one person tested twice)

How to calculate the sign test:

1. State the alternative and null hypothesis 

2. Represent each pair of data with a plus or minus. Can do this by subtracting the values in condition 1 from condition 2, if the results are negative then give a - , if the results are positive give a + sign. If any scores come out the same, give them a 0. This is used in order to work out the calculated value of S = number of times the least frequent sign occurs.

3. Look up the critical value of S in the table. To do this, need to know N (total number of scores given a sign, not including any with a 0) and whether the test hypothesis is one-tailed (directional) or two-tailed (non-directional)

4. Compare the critical value of S with your calculated value of S- following the rule stated under the table of critical values 

How to report the conclusion:

• If the calculated value is equal to or less than the critical value, the result is significant 

• It can therefore be concluded that (...insert hypothesis) can be accepted/rejected 

Reporting on psychological investigations

1. Abstract= key details of the report 

2. Introduction= Past research on the topic, aims and hypotheses

3. Method= What the researcher did; Design, sample, materials, procedure, ethics

4. Results= Descriptive and inferential statistics

5. Discussion= Summary of the results and what they tells us in terms of theory

6. Referencing= Lists of sources that are referred to in the article

Ethics in a report

Need to include an informed consent form, debrief form and standardised instructions

Standardised instructions

To ensure all participants have the same experience, standardised instructions are read as a script to participants. These are given in the chosen location, and there may be more than one set for different (IV) conditions of the investigation. Also, they should help eliminate confounding variables such as experiment effects. Instructions can be read to the participants or they can read the instructions themselves 

Referencing a report

reference section (at the end of the report). This allows psychologists reading the report to find the same sources. The convection for psychological report referencing is to use Harvard style referencing 

Harvard style referencing

1. Author

2. Year

3. Title of book

4. Edition

5. Place

6. Publisher