The readings that I have completed throughout this week have not only widened my perspectives on the constraints of conventional null hypothesis significance testing (NHST) but also reinforced the significance of effect sizes as an immensely crucial aspect of any research conducted in the field of psychology. In the past, my critical analysis has mostly been based on the position of p-values as the most significant measurement of the significance of a finding. A review of Gravetter and Wallnau (2017), lecture by Andy Field (2005) on effect sizes, and the article by Cohen (1994) has enabled me to realise that statistical significance does not necessarily correspond to practical significance.

A key lesson learnt is that effect sizes are measures of the size of a observed effect regardless of the size of the sample. Although a p-value can tell whether an effect is probable to occur due to chance, it cannot tell whether the effect has real-world implications (Gravetter and Wallnau, 2017). Cohen (1994) argued that NHST is a ritualised exercise so that people often mislead researchers to think that statistical significance must be of scientific importance. Such an error may cause an overvaluation of small but statistically significant effects and in particular situations when the size of a sample is large enough to observe trivial effects.

Field (2005) presented the problem with the examples where the same mean differences gave both significant and non-significant results only depending on the sample size. As this demonstration has shown, NHST is greatly dependent on the sample size and thus a poor single measure of the significance of an effect. By contrast, due to their standardised representation, effect sizes, including those of Cohen d and Pearson r, allow comparison across studies and scales of measurement to enable the accrual of cumulative knowledge and reproducibility in psychology.

The fact that Cohen (1994) states that the null hypothesis is hardly ever true in the behavioural sciences was also particularly enlightening to me. Due to the nature variability of human behaviour, there is no true no-difference between the groups to expect. This then implies that the habitual rejection of the null hypothesis does not provide much informative value and that interests of estimating effect sizes and their confidence intervals are better suited to scientific use.

The development of a finer sense of effect sizes has not just sharpened my ability to respond critically to research results, but also will serve as a beneficial future addition to my own work, be it in the appraisal of clinical interventions, experimental design, or provision of evidence based practise. As an illustration, knowing that a treatment has a statistically significant effect is only of limited use unless one is also aware of and evaluates the size of that effect- what it means in practise to clients or communities.

In the future, I will use this knowledge by systematically searching, computing and reporting effect sizes alongside p-values. By valuing the scale and statistical significance of effects, I hope to be a more responsible practitioner in the field as well as to make sure that psychological science is rigorous and relevant.

References

Cohen, J. (1994). The earth is round (p < .05). American Psychologist, 49(12), 997–1003. https://doi.org/10.1037/0003-066X.49.12.997

Field, A. P. (2005). Effect sizes [Lecture handout]. University of Sussex.

Gravetter, F. J., & Wallnau, L. B. (2017). Statistics for the behavioral sciences (10th ed.). Cengage Learning.

The readings that I have completed throughout this week have not only widened my perspectives on the constraints of conventional null hypothesis significance testing (NHST) but also reinforced the significance of effect sizes as an immensely crucial aspect of any research conducted in the field of psychology. In the past, my critical analysis has mostly been based on the position of p-values as the most significant measurement of the significance of a finding. A review of Gravetter and Wallnau (2017), lecture by Andy Field (2005) on effect sizes, and the article by Cohen (1994) has enabled me to realise that statistical significance does not necessarily correspond to practical significance.

A key lesson learnt is that effect sizes are measures of the size of a observed effect regardless of the size of the sample. Although a p-value can tell whether an effect is probable to occur due to chance, it cannot tell whether the effect has real-world implications (Gravetter and Wallnau, 2017). Cohen (1994) argued that NHST is a ritualised exercise so that people often mislead researchers to think that statistical significance must be of scientific importance. Such an error may cause an overvaluation of small but statistically significant effects and in particular situations when the size of a sample is large enough to observe trivial effects.

Field (2005) presented the problem with the examples where the same mean differences gave both significant and non-significant results only depending on the sample size. As this demonstration has shown, NHST is greatly dependent on the sample size and thus a poor single measure of the significance of an effect. By contrast, due to their standardised representation, effect sizes, including those of Cohen d and Pearson r, allow comparison across studies and scales of measurement to enable the accrual of cumulative knowledge and reproducibility in psychology.

The fact that Cohen (1994) states that the null hypothesis is hardly ever true in the behavioural sciences was also particularly enlightening to me. Due to the nature variability of human behaviour, there is no true no-difference between the groups to expect. This then implies that the habitual rejection of the null hypothesis does not provide much informative value and that interests of estimating effect sizes and their confidence intervals are better suited to scientific use.

The development of a finer sense of effect sizes has not just sharpened my ability to respond critically to research results, but also will serve as a beneficial future addition to my own work, be it in the appraisal of clinical interventions, experimental design, or provision of evidence based practise. As an illustration, knowing that a treatment has a statistically significant effect is only of limited use unless one is also aware of and evaluates the size of that effect- what it means in practise to clients or communities.

In the future, I will use this knowledge by systematically searching, computing and reporting effect sizes alongside p-values. By valuing the scale and statistical significance of effects, I hope to be a more responsible practitioner in the field as well as to make sure that psychological science is rigorous and relevant.

References

Cohen, J. (1994). The earth is round (p < .05). American Psychologist, 49(12), 997–1003. https://doi.org/10.1037/0003-066X.49.12.997

Field, A. P. (2005). Effect sizes [Lecture handout]. University of Sussex.

Gravetter, F. J., & Wallnau, L. B. (2017). Statistics for the behavioral sciences (10th ed.). Cengage Learning.

The readings that I have completed throughout this week have not only widened my perspectives on the constraints of conventional null hypothesis significance testing (NHST) but also reinforced the significance of effect sizes as an immensely crucial aspect of any research conducted in the field of psychology. In the past, my critical analysis has mostly been based on the position of p-values as the most significant measurement of the significance of a finding. A review of Gravetter and Wallnau (2017), lecture by Andy Field (2005) on effect sizes, and the article by Cohen (1994) has enabled me to realise that statistical significance does not necessarily correspond to practical significance.

A key lesson learnt is that effect sizes are measures of the size of a observed effect regardless of the size of the sample. Although a p-value can tell whether an effect is probable to occur due to chance, it cannot tell whether the effect has real-world implications (Gravetter and Wallnau, 2017). Cohen (1994) argued that NHST is a ritualised exercise so that people often mislead researchers to think that statistical significance must be of scientific importance. Such an error may cause an overvaluation of small but statistically significant effects and in particular situations when the size of a sample is large enough to observe trivial effects.

Field (2005) presented the problem with the examples where the same mean differences gave both significant and non-significant results only depending on the sample size. As this demonstration has shown, NHST is greatly dependent on the sample size and thus a poor single measure of the significance of an effect. By contrast, due to their standardised representation, effect sizes, including those of Cohen d and Pearson r, allow comparison across studies and scales of measurement to enable the accrual of cumulative knowledge and reproducibility in psychology.

The fact that Cohen (1994) states that the null hypothesis is hardly ever true in the behavioural sciences was also particularly enlightening to me. Due to the nature variability of human behaviour, there is no true no-difference between the groups to expect. This then implies that the habitual rejection of the null hypothesis does not provide much informative value and that interests of estimating effect sizes and their confidence intervals are better suited to scientific use.

The development of a finer sense of effect sizes has not just sharpened my ability to respond critically to research results, but also will serve as a beneficial future addition to my own work, be it in the appraisal of clinical interventions, experimental design, or provision of evidence based practise. As an illustration, knowing that a treatment has a statistically significant effect is only of limited use unless one is also aware of and evaluates the size of that effect- what it means in practise to clients or communities.

In the future, I will use this knowledge by systematically searching, computing and reporting effect sizes alongside p-values. By valuing the scale and statistical significance of effects, I hope to be a more responsible practitioner in the field as well as to make sure that psychological science is rigorous and relevant.

References

Cohen, J. (1994). The earth is round (p < .05). American Psychologist, 49(12), 997–1003. https://doi.org/10.1037/0003-066X.49.12.997

Field, A. P. (2005). Effect sizes [Lecture handout]. University of Sussex.

Gravetter, F. J., & Wallnau, L. B. (2017). Statistics for the behavioral sciences (10th ed.). Cengage Learning.

The readings that I have completed throughout this week have not only widened my perspectives on the constraints of conventional null hypothesis significance testing (NHST) but also reinforced the significance of effect sizes as an immensely crucial aspect of any research conducted in the field of psychology. In the past, my critical analysis has mostly been based on the position of p-values as the most significant measurement of the significance of a finding. A review of Gravetter and Wallnau (2017), lecture by Andy Field (2005) on effect sizes, and the article by Cohen (1994) has enabled me to realise that statistical significance does not necessarily correspond to practical significance.

A key lesson learnt is that effect sizes are measures of the size of a observed effect regardless of the size of the sample. Although a p-value can tell whether an effect is probable to occur due to chance, it cannot tell whether the effect has real-world implications (Gravetter and Wallnau, 2017). Cohen (1994) argued that NHST is a ritualised exercise so that people often mislead researchers to think that statistical significance must be of scientific importance. Such an error may cause an overvaluation of small but statistically significant effects and in particular situations when the size of a sample is large enough to observe trivial effects.

Field (2005) presented the problem with the examples where the same mean differences gave both significant and non-significant results only depending on the sample size. As this demonstration has shown, NHST is greatly dependent on the sample size and thus a poor single measure of the significance of an effect. By contrast, due to their standardised representation, effect sizes, including those of Cohen d and Pearson r, allow comparison across studies and scales of measurement to enable the accrual of cumulative knowledge and reproducibility in psychology.

The fact that Cohen (1994) states that the null hypothesis is hardly ever true in the behavioural sciences was also particularly enlightening to me. Due to the nature variability of human behaviour, there is no true no-difference between the groups to expect. This then implies that the habitual rejection of the null hypothesis does not provide much informative value and that interests of estimating effect sizes and their confidence intervals are better suited to scientific use.

The development of a finer sense of effect sizes has not just sharpened my ability to respond critically to research results, but also will serve as a beneficial future addition to my own work, be it in the appraisal of clinical interventions, experimental design, or provision of evidence based practise. As an illustration, knowing that a treatment has a statistically significant effect is only of limited use unless one is also aware of and evaluates the size of that effect- what it means in practise to clients or communities.

In the future, I will use this knowledge by systematically searching, computing and reporting effect sizes alongside p-values. By valuing the scale and statistical significance of effects, I hope to be a more responsible practitioner in the field as well as to make sure that psychological science is rigorous and relevant.

References

Cohen, J. (1994). The earth is round (p < .05). American Psychologist, 49(12), 997–1003. https://doi.org/10.1037/0003-066X.49.12.997

Field, A. P. (2005). Effect sizes [Lecture handout]. University of Sussex.

Gravetter, F. J., & Wallnau, L. B. (2017). Statistics for the behavioral sciences (10th ed.). Cengage Learning.

The readings that I have completed throughout this week have not only widened my perspectives on the constraints of conventional null hypothesis significance testing (NHST) but also reinforced the significance of effect sizes as an immensely crucial aspect of any research conducted in the field of psychology. In the past, my critical analysis has mostly been based on the position of p-values as the most significant measurement of the significance of a finding. A review of Gravetter and Wallnau (2017), lecture by Andy Field (2005) on effect sizes, and the article by Cohen (1994) has enabled me to realise that statistical significance does not necessarily correspond to practical significance.

A key lesson learnt is that effect sizes are measures of the size of a observed effect regardless of the size of the sample. Although a p-value can tell whether an effect is probable to occur due to chance, it cannot tell whether the effect has real-world implications (Gravetter and Wallnau, 2017). Cohen (1994) argued that NHST is a ritualised exercise so that people often mislead researchers to think that statistical significance must be of scientific importance. Such an error may cause an overvaluation of small but statistically significant effects and in particular situations when the size of a sample is large enough to observe trivial effects.

Field (2005) presented the problem with the examples where the same mean differences gave both significant and non-significant results only depending on the sample size. As this demonstration has shown, NHST is greatly dependent on the sample size and thus a poor single measure of the significance of an effect. By contrast, due to their standardised representation, effect sizes, including those of Cohen d and Pearson r, allow comparison across studies and scales of measurement to enable the accrual of cumulative knowledge and reproducibility in psychology.

The fact that Cohen (1994) states that the null hypothesis is hardly ever true in the behavioural sciences was also particularly enlightening to me. Due to the nature variability of human behaviour, there is no true no-difference between the groups to expect. This then implies that the habitual rejection of the null hypothesis does not provide much informative value and that interests of estimating effect sizes and their confidence intervals are better suited to scientific use.

The development of a finer sense of effect sizes has not just sharpened my ability to respond critically to research results, but also will serve as a beneficial future addition to my own work, be it in the appraisal of clinical interventions, experimental design, or provision of evidence based practise. As an illustration, knowing that a treatment has a statistically significant effect is only of limited use unless one is also aware of and evaluates the size of that effect- what it means in practise to clients or communities.

In the future, I will use this knowledge by systematically searching, computing and reporting effect sizes alongside p-values. By valuing the scale and statistical significance of effects, I hope to be a more responsible practitioner in the field as well as to make sure that psychological science is rigorous and relevant.

References

Cohen, J. (1994). The earth is round (p < .05). American Psychologist, 49(12), 997–1003. https://doi.org/10.1037/0003-066X.49.12.997

Field, A. P. (2005). Effect sizes [Lecture handout]. University of Sussex.

Gravetter, F. J., & Wallnau, L. B. (2017). Statistics for the behavioral sciences (10th ed.). Cengage Learning.

The readings that I have completed throughout this week have not only widened my perspectives on the constraints of conventional null hypothesis significance testing (NHST) but also reinforced the significance of effect sizes as an immensely crucial aspect of any research conducted in the field of psychology. In the past, my critical analysis has mostly been based on the position of p-values as the most significant measurement of the significance of a finding. A review of Gravetter and Wallnau (2017), lecture by Andy Field (2005) on effect sizes, and the article by Cohen (1994) has enabled me to realise that statistical significance does not necessarily correspond to practical significance.

A key lesson learnt is that effect sizes are measures of the size of a observed effect regardless of the size of the sample. Although a p-value can tell whether an effect is probable to occur due to chance, it cannot tell whether the effect has real-world implications (Gravetter and Wallnau, 2017). Cohen (1994) argued that NHST is a ritualised exercise so that people often mislead researchers to think that statistical significance must be of scientific importance. Such an error may cause an overvaluation of small but statistically significant effects and in particular situations when the size of a sample is large enough to observe trivial effects.

Field (2005) presented the problem with the examples where the same mean differences gave both significant and non-significant results only depending on the sample size. As this demonstration has shown, NHST is greatly dependent on the sample size and thus a poor single measure of the significance of an effect. By contrast, due to their standardised representation, effect sizes, including those of Cohen d and Pearson r, allow comparison across studies and scales of measurement to enable the accrual of cumulative knowledge and reproducibility in psychology.

The fact that Cohen (1994) states that the null hypothesis is hardly ever true in the behavioural sciences was also particularly enlightening to me. Due to the nature variability of human behaviour, there is no true no-difference between the groups to expect. This then implies that the habitual rejection of the null hypothesis does not provide much informative value and that interests of estimating effect sizes and their confidence intervals are better suited to scientific use.

The development of a finer sense of effect sizes has not just sharpened my ability to respond critically to research results, but also will serve as a beneficial future addition to my own work, be it in the appraisal of clinical interventions, experimental design, or provision of evidence based practise. As an illustration, knowing that a treatment has a statistically significant effect is only of limited use unless one is also aware of and evaluates the size of that effect- what it means in practise to clients or communities.

In the future, I will use this knowledge by systematically searching, computing and reporting effect sizes alongside p-values. By valuing the scale and statistical significance of effects, I hope to be a more responsible practitioner in the field as well as to make sure that psychological science is rigorous and relevant.

References

Cohen, J. (1994). The earth is round (p < .05). American Psychologist, 49(12), 997–1003. https://doi.org/10.1037/0003-066X.49.12.997

Field, A. P. (2005). Effect sizes [Lecture handout]. University of Sussex.

Gravetter, F. J., & Wallnau, L. B. (2017). Statistics for the behavioral sciences (10th ed.). Cengage Learning.

The readings that I have completed throughout this week have not only widened my perspectives on the constraints of conventional null hypothesis significance testing (NHST) but also reinforced the significance of effect sizes as an immensely crucial aspect of any research conducted in the field of psychology. In the past, my critical analysis has mostly been based on the position of p-values as the most significant measurement of the significance of a finding. A review of Gravetter and Wallnau (2017), lecture by Andy Field (2005) on effect sizes, and the article by Cohen (1994) has enabled me to realise that statistical significance does not necessarily correspond to practical significance.

A key lesson learnt is that effect sizes are measures of the size of a observed effect regardless of the size of the sample. Although a p-value can tell whether an effect is probable to occur due to chance, it cannot tell whether the effect has real-world implications (Gravetter and Wallnau, 2017). Cohen (1994) argued that NHST is a ritualised exercise so that people often mislead researchers to think that statistical significance must be of scientific importance. Such an error may cause an overvaluation of small but statistically significant effects and in particular situations when the size of a sample is large enough to observe trivial effects.

Field (2005) presented the problem with the examples where the same mean differences gave both significant and non-significant results only depending on the sample size. As this demonstration has shown, NHST is greatly dependent on the sample size and thus a poor single measure of the significance of an effect. By contrast, due to their standardised representation, effect sizes, including those of Cohen d and Pearson r, allow comparison across studies and scales of measurement to enable the accrual of cumulative knowledge and reproducibility in psychology.

The fact that Cohen (1994) states that the null hypothesis is hardly ever true in the behavioural sciences was also particularly enlightening to me. Due to the nature variability of human behaviour, there is no true no-difference between the groups to expect. This then implies that the habitual rejection of the null hypothesis does not provide much informative value and that interests of estimating effect sizes and their confidence intervals are better suited to scientific use.

The development of a finer sense of effect sizes has not just sharpened my ability to respond critically to research results, but also will serve as a beneficial future addition to my own work, be it in the appraisal of clinical interventions, experimental design, or provision of evidence based practise. As an illustration, knowing that a treatment has a statistically significant effect is only of limited use unless one is also aware of and evaluates the size of that effect- what it means in practise to clients or communities.

In the future, I will use this knowledge by systematically searching, computing and reporting effect sizes alongside p-values. By valuing the scale and statistical significance of effects, I hope to be a more responsible practitioner in the field as well as to make sure that psychological science is rigorous and relevant.

References

Cohen, J. (1994). The earth is round (p < .05). American Psychologist, 49(12), 997–1003. https://doi.org/10.1037/0003-066X.49.12.997

Field, A. P. (2005). Effect sizes [Lecture handout]. University of Sussex.

Gravetter, F. J., & Wallnau, L. B. (2017). Statistics for the behavioral sciences (10th ed.). Cengage Learning.