Choosing and how to do a statistical test

What level of measurement is the data?

Collected Data

Nominal

Is the data a test of association/correlation or a test of difference and if so are they related

Create an aim and hypothesis and conduct an experiment

Unrelated Design

Chi-squared

Related Design

Sign test

Correlation / Association

Chi-squared

Ordinal

Is the data a test of association/correlation or a test of difference and if so are they related

Unrelated Design

Mann-Whitney

Related Design

Wilcoxon

Correlation / Association

Spearman's rho

Interval

Is the data a test of association/correlation or a test of difference and if so are they related

Unrelated Design

Unrelated t-test

Related Design

Related t-test

Correlation / Association

Pearson's r

Step 1

Rank the rating considering both groups and rating them together. Lowest number has a rank of 1. If you have more than 1 of the same number use the mean of the ranks so 8 and 9 will be 8.5

Step 2

Work out the value of U using one of:

- UA = RA - [NA(NA + 1]

- UB = RB - [NB(NB + 1)]

or

- Ua = nanb + {na(na+1)}/{2} - E`Ra

- Ub = nanb + {nb(nb+1)}/{2} - E`Ra

Step 3:

Work out the critical value and if the results were significant, the calculated value has to be less than or equal to the critical value

Step 1

Calculate a difference and rank the difference. When ranking ignore the signs. If the difference is zero the data is not included in the ranking

Step 2

Working out the value of T. T is the calculated value which is the sum of the less frequent sign

Step 3

Find the crtical value and work out if the results are significant. To by significant the calculated T value must br less than or equal to the critical value

Step 1

Rank each set of scores separately in each condition from lowest to highesst. If there are two of the same score take the mean of the ranks

Step 2

Find the difference between each pair of ranks and square the difference. Add up the difference of the ranks

Step 3

Work out the value of rho using:

rho = {6E`d^2}/{N(N^{2} -1}

Step 4

Find out the critical value and see if the results are significant. To be significant the calculated value (correlation value) must be equal to or more than the critical value

Step 1

Calculate the sum of the scores for x and y

Step 2

Square each x and value and add up the values

Step 3

Multiply x and y for each participant and add up these values

Step 4

Work out the value of r using:

r = {N(E`xy) - (E`x)(E`y)}/{¬{[NE`x^{2} - (E¬x)^{2}][NE`y^{2} - (E`y)^{2}}

Step 5

Calculate the df value by doing N - 2 and finding the critical value. For the results to be significant the calculated value (r) must be bigger than or equal to the critical value

Step 1

Calculate the sum of the scores for group A and B

Step 2

Square each x and value and add up the values

Step 3

Work out the values for Sa = {E`Xa^{2} - (E`Xa)^{2}}/{Na}

Sb = {E`Xb^{2} - (E`Xa)^{2}}/{Nb}

Step 4

Work out the value of t by doing:

t = {X`a - X`b}/{ ¬{({Sa + Sb}/{Na + Nb - 2})({Na + Nb}/{NaNb})} }

Step 5

Work out the value of df by doing Na + Nb - 2. Find the critical value and see if the results are significant. To be significant the calculated value of t must be equal to or more than the critical value

Step 1

Calculate the difference (d) between scores for condition A and condition B. Then add up the values

Step 2

Square each difference and add up the values

Step 3

Work out the value of t by doing:

t = ¬{{E`d^{2} - {E`d)^{2}}/{N}}/{N(N - 1)}

Step 3

Work out the value of df by doing N - 1 and find out the critical value. To be significant the calculated value must be equal to or more than the critical value.

Step 1

Draw out a a contingency table

Step 2

Calculate the expected frequencies fro all of the cells. An expected value if the value expected if there was no difference between the two groups

Step 3

Work out the value of E - O for each cell

Step 4

Square the value for E -O and then divid the value by E. This is the X^2 value (chi-squared value)

Step 5

Calculate the df value by doing (rows - 1) * (columns -1). Find out the critical value and work out if the results are significant. The value if significant if the X^2 value (calculated value) must be equal to or more than the critical value

Step 1

Draw out a a contingency table

Step 2

Calculate the expected frequencies fro all of the cells. An expected value if the value expected if there was no difference between the two groups

Step 3

Work out the value of E - O for each cell

Step 4

Square the value for E -O and then divid the value by E. This is the X^2 value (chi-squared value)

Step 5

Calculate the df value by doing (rows - 1) * (columns -1). Find out the critical value and work out if the results are significant. The value if significant if the X^2 value (calculated value) must be equal to or more than the critical value

Step 1

Subtact the one score from the other but only record the sign so if the answer is negative recod a - sign or if it is positive recod a + sign. If the answer is 0 record a = sign

Step 2

Add up all the positive and negative sign

Step 3

Take the less frequent sign, this is the calculated value.

Step 4

Find out the critical value and find out if the results are significant. To be significant the calculated value must be equal to or less than the critical value