Presentasi sedang didownload. Silahkan tunggu

Presentasi sedang didownload. Silahkan tunggu

Research methodology and Scientific Writing W#8

Presentasi serupa


Presentasi berjudul: "Research methodology and Scientific Writing W#8"— Transcript presentasi:

1 Statistics for IS/IT Research – 2 Betty Purwandari, PhD Dana Indra Sensuse, PhD
Research methodology and Scientific Writing W#8 Faculty of Computer Science University of Indonesia 2017

2 Review: Statistik Deskriptif dan Inferensial
Digunakan untuk mendeskripsikan sampel, dan tidak ingin membuat kesimpulan yang berlaku untuk populasi. Statistik Inferensial: Menggunakan sampel untuk membuat inferensi yang berlaku untuk populasi  melakukan generalisasi. Ada uji signifikansi dan taraf kesalahan.

3 Conduct Hypothesis Tests
Identify a null and alternative hypothesis Set the level of significance (α = alpha level) for rejecting the null hypothesis Collect data Compute the sample statistic Make a decision about rejecting/failing to reject

4 Contoh Hipotesis Rumusan MasalahDeskriptif: Hipotesis:
Seberapa baik pelayanan IT di perusahaan X? Hipotesis: Pelayanan IT di perusahaan X paling tinggi 80% dari yang diharapkan Hipotesis Operasional: Ho: μ <= 80% H1: μ > 80%

5 Contoh Hipotesis Rumusan Masalah Asosiatif: Hipotesis: Hipotesis Nol:
Adakah hubungan yang positif dan signifikan antara kualitas pelayanan TI dengan kinerja perusahaan? Hipotesis: Diduga terdapat hubungan yang positif dan signifikan antara kualitas pelayanan TI dengan kinerja perusahaan. Hipotesis Nol: Ho: Tidak ada hubungan yang positif dan signifikan antara kualitas pelayanan TI dengan kinerja perusahaan.

6 Contoh Hipotesis Rumusan Masalah Komparatif: Hipotesis: Hipotesis Nol:
Adakah perbedaan yang signifikan pada layanan TI antara cabang A, B , dan C? Hipotesis: Terdapat perbedaan yang signifikan pada layanan TI antara cabang A, B, dan C. Hipotesis Nol: Ho: Tidak ada perbedaan yang signifikan pada layanan TI antara cabang A, B, dan C.

7 Membuat Keputusan untuk Menolak atau Menerima Hipotesis Nol
Ho ditolak: Penelitian tersebut terbukti secara nyata (empiris). Ho diterima: Penelitian tersebut tidak nyata secara empiris.

8 Level of Significance The significance testing is a blend of:
Fisher’s idea of using the probability value p as an index of the weight of evidence against a null hypothesis Jerzy Neyman and Egron Pearson’s idea of testing a null hypothesis against an alternative hypothesis. Fisher suggested that 95% is a useful threshold for confidence We are 95% certain that a result is genuine (not by chance) Or there is only a 5% chance (a probability of 0.05) of something occurring by chance

9 Level of Significance Report summaries with:
One star (*) attached to indicate p <= 0.05 For IS/IT research Two stars (**) to indicate p <= 0.01 Occasionally, three stars (***) were used to indicate p <= For research in medicine

10 Type I Error (α = Alpha Level)
The incorrect rejection of a true null hypothesis. Conclude that a supposed effect or relationship exists when in fact it does not. The researchers believe that there is a genuine effect in the population, when in fact there is NOT. Disimpulkan bhw TERDAPAT perbedaan yang signifikan pada layanan TI antara cabang A, B, dan C. Padahal sebenarnya TIDAK TERDAPAT perbedaan yang signifikan. Ho ditolak, padahal Ho benar. Fisher’s criterion: The probability of this error is .05 (or 5%) when there is no effect in the population. Assuming there is no effect in our population, if the researchers replicated our data collection 100 times: On five occasions we would obtain a test statistic large enough to make us think that there was a genuine effect in the population even though there is NOT.

11 Type II Error (β = Beta Level)
The failure to reject a false null hypothesis. The researchers believe that there is NO effect in the population when, in reality, there IS. Disimpulkan bhw TIDAK TERDAPAT perbedaan yang signifikan pada layanan TI antara cabang A, B, dan C. Padahal sebenarnya TERDAPAT perbedaan yang signifikan. Ho diterima, padahal Ho salah. The maximum acceptable probability of a Type II error would be 0.2 (or 20%). The power or sensitivity is equal to 1−β. β = 20%, power = 80% If the researchers took 100 samples of data from a population in which an effect exists, they would fail to detect that effect in 20 of those samples (so they would miss 1 in 5 genuine effects).

12 Variance Measures how far a set of number is spread out.
A variance of zero indicates that all the values are identical. Using the error in the sample to estimate the error in the population. SS = Sum of Squared Errors = sample mean N = sample size s = standard deviation

13 Effect Size The size of an effect (be that an experimental manipulation or the strength of a relationship between variables). A correlation coefficient of 0 means there is no effect, and a value of 1 means that there is a perfect effect. A small, medium or large effect as percentage of total variance.

14 Hubungan versus Pengaruh
Pengaruh lamanya penayangan iklan di TV terhadap nilai penjualan barang: X  Y Rumusan masalah: Berapakah rata-rata waktu penanyangan iklan di TV? Berapakah nilai penjualan barang yang telah diiklankan? Apakah ada hubungan positif dan signifikan antara lamanya penayangan iklan di TV dengan nilai penjualan barang? Bagaimana pengaruh lama penayangan iklan di TV terhadap nilai penjualan barang?

15 Assumptions for Using Parametric Tests

16 Advantages of Nonparametric Tests

17 Choosing Appropriate Statistical Technique

18

19 Testing Hypothesis on a Single Mean
One sample t-test: To test the hypothesis that the mean of the population from which a sample is drawn is equal to a comparison standard. Seberapa tinggi produktifitas pegawai di PT XYZ?

20 Testing Hypotheses about Two Related Means
Paired samples t-test: A parametric test Examines differences in the same group before and after a treatment. The Wilcoxon signed-rank test: A non parametric test for examining significant differences between two related samples or repeated measurement on a single sample. Used when the population cannot be assumed to be normally distributed Ordinal data. Apakah ada perbedaan IP mahasiswa sebelum dan sesudah menggunakan e-learning?

21 Testing Hypotheses about Two Related Means
McNemar’s test: Non parametric test on nominal data. It assesses the significance of the difference between two dependent samples when the variable of interest is dichotomous. It is used primarily in before-after studies to test for an experimental effect.

22 Testing Hypotheses about Two Unrelated Means
Independent samples t-test: To see if there are any significant differences in the means for two groups in the variable of interest. Apakah ada perbedaan IPK antara pria dan wanita?

23 How to Calculate Sample Size
G Power: a tool to calculate sample size Available on Compute required sample size: given α , power (1−β), and effect size.

24 An Example T-test: testing hypotheses about two unrelated (independent) means α = 5% Power (1−β) = 80% Effect size = d = 0.5 (medium effect size) The effect accounts for 50% of the variance

25 Testing Hypotheses about Several Means
ANalysis of VAriance (ANOVA): Examine the significant mean differences among more than two groups on an interval or ratio-scaled dependent variable.

26 Regression Analysis Simple regression: Multiple regression analysis:
One metric independent variable is hypothesized to affect one metric dependent variable. Multiple regression analysis: More than one (metric or non-metric) independent variable to explain variance in a (metric) dependent variable.

27 Multivariate Analysis
Metode analisis statistik yang melibatkan multi-variable secara simultan

28 Multivariate Analysis
Contoh: Structural Equation Modeling (SEM) How big is the sample for SEM? Min sample size 100 to 500, depends on the number of constructs (unobserved variables) and communalities (how good the unobserved constructs explaining variance in the measured variables).

29 An Example of SEM

30 References Field, A., Discovering Statistics Using SPSS, 3rd ed. SAGE Publications Ltd, London. Hair, J.F., Black, W.C., Babin, B.J., Anderson, R.E., Multivariate Data Analysis, 7th ed. Pearson Prentice Hall Publishing.


Download ppt "Research methodology and Scientific Writing W#8"

Presentasi serupa


Iklan oleh Google