# Taping in data

Descargar 23.1 Kb.
 Fecha de conversión 04.05.2017 Tamaño 23.1 Kb.
 TAPING IN DATA Each line represents a participant. In transversal studies, each column is a different variable. In longitudinal studies, each column is a condition of the independent variable. In “Data View” (tab in down left corner), only numbers can be included. In “Variable View” (tab in down left corner), each line is a variable. It is important to determine the measure of each variable (in last column, nominal, ordinal o scale). CONVERSION FROM QUANTITATIVE TO QUALITATIVE OR ORDINAL VARIABLE Transformar – recodificar en diferentes variables – introducir la variable a transformar – poner otro nombre – antiguos y nuevos valores – (you have the possibility of choosing ranges – ELEGIR-) STANDARD SCORES Analyze – Descriptive statistics – descriptive – click “save standardized values as variable” In order to sum or subtract values of two variables in each participant: Transformar – calcular variable ASSUMPTIONS 1. NORMALITY (in the quantitative variable/s, test normal distribution): Analyze – nonparametric tests – 1 sample K-S When Kolmogorov Smirnov Z presents sig ≥ 0.05, the distribution is normal 2. HOMOSCEDASTICITY (in the relationship between qualitative/ordinal – quantitative variables) Analyze – Compare Means – One-Way ANOVA – Options – Homogeneity of variance test When Levene statistic presents sig ≥ 0.05, there is homoscedasticity 3. LINEARITY (in the relationship between ordinal/quantitative (with more than two groups – quantitative variables) Analyze – Compare Means – Means – Options – Test for linearity sig < 0.05 – There is linearity 4. INDEPENDENCE OF ERRORS (in the relationship between variables) Analyze – Regression – Linear – Statistics – Durbin-Watson 1.5 < d < 2.5 – Errors are independent Note in paired samples: you only need to take into account the sphericity (assumptions 1-4 have not to be checked). GRAPHS Relationship between two quantitative variables – Graphs - Legacy Dialogs -Scatter/Dot Relationship between a qualitative and a quantitative variable – Graphs - Legacy Dialogs - Bar STATISTICS 0. Compare values in the sample and the population of reference. Comparar medias – prueba T para una media. In “valor para una media”, include the value of the population. Sig < 0.05 – The sample comes from a different population. 1. CHI SQUARE TEST Analyze – Descriptive statistics –crosstabs Include one variable in ROWS and the other in COLUMNS Statistics – click in Chi-square – OK Sig < 0.05 – relationship between variables / differences across groups. 2. FISHER EXACT TEST Analyze – Descriptive statistics –crosstabs Include one variable in ROWS and the other in COLUMNS Statistics – click in Chi-square – OK Fisher’s Exact Test Sig < 0.05 – relationship between variables / differences across groups. 3. MCNEMAR TEST Analyze – Descriptive statistics –crosstabs Include one variable in ROWS and the other in COLUMNS Statistics – click in McNemar – OK Sig < 0.05 – relationship between variables / differences across groups. 4. MANN WHITNEY U Analyze – Nonparametric Tests – Two Independent Samples – Mann-Whitney U Sig < 0.05 – relationship between variables / differences across groups. 5. KRUSKAL WALLIS H Analyze – Nonparametric Tests – k Independent Samples – Kruskal-Wallis H Sig < 0.05 – relationship between variables / differences across groups. 6. WILCOXON T TEST Analyze – Nonparametric Tests – Two Related Samples - Wilcoxon Sig < 0.05 – relationship between variables / differences across groups. 7. T (INDEPENDENT SAMPLES): Analyze – Compare Means – Independent-Samples T test Sig < 0.05 – relationship between variables / differences across groups. 8. T (PAIRED SAMPLES): Analyze – Compare Means – Paired-Samples T test Sig < 0.05 – relationship between variables / differences across groups. 9. F (TRANSVERSAL): Analyze – compare means – one-way ANOVA – Sig < 0.05 – relationship between variables / differences across groups. 9.1. POST_HOC CONTRASTS: post-hoc – Scheffé, Tukey 9.2. A PRIORI CONTRASTS: contrasts – coefficients – add 9.2.1. In tendency contrasts: + polynomial 10. F (LONGITUDINAL): Analyze – general linear model – repeated measures Greenhouse-Geisser Sig < 0.05 – relationship between variables / differences across groups. 11. F (MIXED DESIGNS): Analizar – modelo lineal general – medidas repetidas. The factor within groups (intrasujetos) is the longitudinal variable. Its number of levels is its number of measures. The factor between groups (intersujetos) is the transversal variable. Sig < 0.05: relationship between variables / differences across groups: sig of the between groups test (intersujetos) (the last table that appears) to see if there are differences in the conditions of the transversal variable In the within groups test (intrasujetos): Sig of factor 1 (esfericity) to see if there are differences in the longitudinal variable. The most interesting: sig of factor 1 * the transversal variable to see if there are differences in the interaction. 12. PEARSON CORRELATION: Analyze - correlate - bivariate - Pearson Sig < 0.05: relationship between variables 13. SPEARMAN CORRELATION: Analyze - correlate - bivariate - Spearman Sig < 0.05: relationship between variables 14. SIMPLE LINEAR REGRESSION: Analize – regression – linear Sig < 0.05: relationship between variables

La base de datos está protegida por derechos de autor ©bazica.org 2016
enviar mensaje